When detecting a license report if a full license text, or only a notice or mention was detected

[pombredanne]

Knowing if we detected a whole license text, or only a notice or mention is a relevant data to later bring some automated summarization and reasoning on what was detected.

For instance, say you have a directory with a COPYING file at the root with a full LGPL 2.1 license text and that all or most source files have a simple notice in a comment Licensed under the LGPL and that one file carries a BSD text.

With this data in hand, you could conclude that the code is overall LGPL-2.1 licensed inferring the version from the top level full text and that one file is BSD-licensed .

To implement this there are a possible avenues:

1. automated using the size of a match and comparing that to the size of a corresponding full license text:
   • if matched to a full text or if bigger or equal to the license text size: this is a full text match
   • a small match of a few line significantly smaller than than the text, this could be a mention, such as licensed under the same license as Perl or license: BSD .
   • a medium match of a several lines and still smaller than than the full text, this could be a notice
2. data driven using tagged rules
   we would tag each rule YAML as a full text, notice or mention and report that

The two approach can be combined and we could start with 1. and progressively adopt 2. (that requires updating several thousands rules... ;) )

[pombredanne]

We will implement reporting a confidence of a match in #342 so I am not sure this is still valid, but leaving this open for now.

[pombredanne]

The develop branch now reports each license with flags telling if this is a full text, notice, etc